Hello and thank you for making such a great product.
I need a little bit strange configuration of Proxmox cluster:
1. I need three nodes cluster;
2. node 1 and node 2 have the same hardware configuration and are not used by now (no storage/vms).
3. node 3 is running in production with multiple VMs, some VMs will be migrated to nodes 1 and 2;
4. node 1 and node 2 should implement shared DRBD8 storage, exported as iSCSI target, to be available on all nodes.
DRBD works in active/passive mode, so from DRBD point of view one of two nodes is primary and provides iSCSCI target via floating IP and other is secondary. From Proxmox point of view nodes 1 and 2 are running VM from iSCSI shared storage, with HA enabled.
I understand that it's unsupported configuration, but can not use separate storage cluster because of some technical/organizational constraints
Active/passive DRBD storage is implemented now on prototype setup using pacemaker (works quite well). Pacemaker uses corosync, as well as Proxmox.
So is it correct sequence to setup cluster starting with third node, that is running in production already:
1. Create cluster on node 3 (with running VMs);
2. Add another two nodes (without VMs) to cluster;
3. Install pacemaker on node 1 and node 2 and setup pacemaker resources for DRBD/IP/iSCSI using crm shell etc.
Is it safe and are there any possible problems/conflicts with corosync with it?
Is it ok to use pacemaker on proxmox cluster for functionality not implemented in Proxmox?
I need a little bit strange configuration of Proxmox cluster:
1. I need three nodes cluster;
2. node 1 and node 2 have the same hardware configuration and are not used by now (no storage/vms).
3. node 3 is running in production with multiple VMs, some VMs will be migrated to nodes 1 and 2;
4. node 1 and node 2 should implement shared DRBD8 storage, exported as iSCSI target, to be available on all nodes.
DRBD works in active/passive mode, so from DRBD point of view one of two nodes is primary and provides iSCSCI target via floating IP and other is secondary. From Proxmox point of view nodes 1 and 2 are running VM from iSCSI shared storage, with HA enabled.
I understand that it's unsupported configuration, but can not use separate storage cluster because of some technical/organizational constraints
Active/passive DRBD storage is implemented now on prototype setup using pacemaker (works quite well). Pacemaker uses corosync, as well as Proxmox.
So is it correct sequence to setup cluster starting with third node, that is running in production already:
1. Create cluster on node 3 (with running VMs);
2. Add another two nodes (without VMs) to cluster;
3. Install pacemaker on node 1 and node 2 and setup pacemaker resources for DRBD/IP/iSCSI using crm shell etc.
Is it safe and are there any possible problems/conflicts with corosync with it?
Is it ok to use pacemaker on proxmox cluster for functionality not implemented in Proxmox?